Tuesday April 26th 2005 London Tier 2 Technical Meeting
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Present
^^^^^^^
IC-HEP:  Owen Maroney, David Colling, Mona Aggarwal
IC-LeSC: Keith Sephton, David McBride,Steve Macgough
RHUL:    -
QMUL:    Alex Martin, Kondo Gnanvo
UCL:     Mark Dixon, Ben Waugh

Apologies from BRUNEL: Paul Kyberd

Agenda
^^^^^^

1. Review of actions from previous meeting.
2. News from GridPP
3. News from LCG/EGEE
4. Site status review: problems and ways forward.
   - Brunel
   - Imperial HEP
   - Imperial LeSc
   - QMUL
   - RHUL
   - UCL
5.AOB


1. Review of actions
^^^^^^^^^^^^^^^^^^^^


>> Submit savannah bug report regarding passing queue parameters to the  PBS scheduler.Alex to write this up.
Status: ongoing


>> SunGrid Engine

 Community of interest in process of being established, after interest 
from a number of sites at EGEE conference.  A mailing list has been
requested.  LeSC work ongoing.  A CE and SGE queue are installed but
are having firewall problems.
  Status: ongoing

>> LCG2 SRPMS

  Chris has supplied scripts to Owen to check existence/availability of 
  SRPMS of a release of LCG.  Owen to test LCG 2_4_0
  Status: ongoing

>> Owen to add FAQ giving recipe for removing job records from PBS server. 
   PBS sometimes thinks a job is still running that has died.
   Status: open

>> Owen to provide recipe to advertise SE as volatile.
   Status: circulated to LT2 mailing list.  Add to Wikki/FAQ.

>> Owen to circulate recipe for using a single MON box for
a site with more than 1CE, and add recipe to FAQ.  This action
was initially for UCL who were short of a MON box for the UCL-HEP
farm.  No longer the need from UCL but sites agreed was still worth
circulating.
  Status: open

2. News from GridPP 
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

DC talked about Service Chanllenge 3 functionality and related VOs specific experiment issues. IC is installing dCache.  Other sites are welcome to start installing SRMs as well.

*Action* DC to provide to sites documents on experiments requirements
for service challenge.

3. News from LCG/EGEE 
^^^^^^^^^^^^^^^^^^^^^^^^^^^^

There was much complaints that the LCG-2_4_0 release date was announced a week earlier then the actual release. The new releases should be easier to upgrade and on time. There was a discussion regarding site critical test and schedule downtime. The new release is due in July.

OM reported at the EGEE conference, David Kelsey and Linda Cornwall
presented a plan for dealing with security vulnerabilities following
complaints from sysadmins and the Deployment Team in the UK.  Newly
discovered vulnerabilities should be reported to a core group who
would be responsible for developing a fix within 2-3 weeks or would
notify all site-admins of the problem.

Vulnerabilities were different from exploits.  In the case of an actual
exploit all sites would be notified immediately.

There was a lot of interest at EGEE in the SGE work.  A mailing list
was requested.  Also gLite developers were interested in the jobmanager
(gLite does not use the globus jobmanager for batch systems but
uses CondorC instead).  LCG have plans for an integration testbed
involving sites outside CERN and IC should be such a site for SGE
integration.

DC mentioned gLite 1.1 is already out.  gLite components should be in the
July release.

*Action* OM to create Community of Interest mailing list for SGE and LCG.

In LCG-2_4_0 BDII port change from 2135 to 2170 and MDS_VO_name is equal to mds-site-name. This change needs to be notified to LCG ROLLOUT and manually 
in GOC database. 

4. Site status review: problems and ways forward.
   - Brunel
        - Upgrade to LCG-2_4_0 on RH7.3 successful. Work ongoing on
SL.
   - Imperial HEP
        - Upgrade to LCG-2_4_0 on RHEL successful. 
   - Imperial LeSc
        - Installation of SGE going on. Some firewall issues.
   - QMUL
        - Upgrade incomplete to LCG-2_4_0 on Fedora.
        - Chris Williams departure has caused problems. Owen will visit QMUL to assist.
   - RHUL
        - Upgrade to LCG-2_4_0 on RHEL successful.
   - UCL
        - Upgrade to LCG-2_4_0 on SL successful.
        - Facing problems in installations of MON_BOX. 
        - job_list_match failures with UCL-HEP
        

5. AOB
^^^^^^
BW asked if there was firm plans to change the site names to all start
with LT2-* or LONDON-*.  General preferment was LONDON-*.

OM said there is a proposal that all UK/Ireland sites should start with UK/I.

*Action* OM to circulate proposal for change.